Capability Antecedents and Performance Outcomes

Capability Antecedents and Performance Outcomes of Servitization: Differences between Basic and Advanced Services

Rui Sousa (Corresponding Author) Professor of Operations Management Universidade Católica Portuguesa, Católica Porto Business School and CEGE

Giovani J.C. da Silveira Professor of Operations and Supply Chain Management Haskayne School of Business University of Calgary

Sousa, R., da Silveira, G. (2017), “Capability antecedents and performance outcomes of servitization: Differences between basic and advanced services”. International Journal of Operations & Production Management, 37(4), 444 – 467.

Acknowledgments: This study was funded by the University of Calgary Distinguished Visiting Research Fellow Program and a Haskayne Research Professorship in Operations and Supply Chain Management. 1

Capability Antecedents and Performance Outcomes of Servitization: Differences between Basic and Advanced Services Structured Abstract Purpose This study theoretically articulates and empirically tests an integrated model of capability antecedents and performance outcomes of servitization strategies. We characterize servitization strategies based on the offering of two types of services: Basic Services (BAS) and Advanced Services (ADS). Methodology Hypotheses are tested based on statistical analyses of a large survey of manufacturers from different countries and sectors. Findings We find that: (i) manufacturing capabilities associate with the provision of BAS, while service capabilities associate with both BAS and ADS; (ii) BAS do not impact financial performance, but support the offering of ADS; (iii) there seem to be naturally occurring servitization trajectories involving the gradual development of balanced levels of BAS and ADS and adequate levels of manufacturing and service capabilities. Limitations The findings on servitization trajectories are based on the observation of manufacturing business units at different stages of servitization (cross-sectional data). Practical implications Manufacturers wishing to servitize should distinguish between BAS and ADS and deploy a balanced adoption of BAS and ADS, using BAS as a platform. This should be accompanied with the building of appropriate capabilities. Originality This is one of the first studies to show an explicit link between different servitization strategies, capabilities and servitization maturity. It provides new insights into the servitization paradox and servitization trajectories. Keywords: Servitization, Manufacturing Strategy, Capabilities, Financial Performance. 2

1. Introduction Manufacturing units have increasingly adopted servitization strategies, competing through product-service systems rather than products alone (Neely, 2008; Baines and Lightfoot, 2014). Despite this trend, empirical studies have raised questions about the impact of servitization on firm financial performance (Fang et al., 2008; Neely, 2008; Kastalli and van Looy, 2013; Kohtamaki et al., 2013; Suarez et al., 2013; Eggert et al., 2014), revealing a “servitization paradox” (Gebauer et al., 2005). They suggest that servitization may have a neutral or negative impact on performance at early stages of offering, after which the impact turns positive. A key reason put forward for the paradox is the investment and implementation challenges associated with the need to build new capabilities (e.g., Neu and Brown, 2005; Galbraith, 2002; Miller et al., 2002; Brax, 2005; Martinez et al., 2010). However, our understanding of the capability antecedents and performance outcomes of servitization is still much incomplete (Gebauer et al., 2012). First, although capability investment can explain performance decline after servitization, few empirical studies have addressed the capability antecedents and performance outcomes of servitization in an integrated model. Second, empirical research has not clearly distinguished between different types of servitization strategies, e.g., based on basic or advanced service offerings (Gebauer et al., 2005) (an exception is Eggert et al., 2014). This distinction is important because different services may require different capabilities and have different performance impacts. Third, different servitization trajectories have been proposed (e.g., from basic to advanced services; Mathieu, 2001b; Oliva and Kallenberg, 2003; Gebauer et al., 2005; Martinez et al. 2010; Kowalkowski et al., 2015), but there has been insufficient examination via large-scale empirical studies of their

3

occurrence and impact on performance. Finally, there is a dearth of evidence on antecedents and performance outcomes of servitization based on large-scale studies (Gebauer et al., 2012; Baines and Lightfoot, 2014; Eloranta and Turunen, 2015). Except for Neely (2008), the few available studies are drawn on relatively narrow samples focusing on one firm (Kastalli and van Looy, 2013), one sector (Kohtamaki et al., 2013; Suarez et al., 2013; Eggert et al., 2014) or one country (Fang et al., 2008; Kohtamaki et al., 2013; Eggert et al., 2014). It is important to consider international samples from diverse sectors because differences in operational environments influence how to servitize (Turunen and Finne, 2014). In order to address these research gaps, our study theoretically articulates and empirically tests an integrated, fine-grained model of capability antecedents and performance outcomes of servitization. We characterize servitization strategies based on the offering of Basic Services (BAS) and Advanced Services (ADS). BAS aim to install and maintain basic product functionality. ADS relate to working closely with customers to co-create value that goes beyond basic product operation, involving the adaptation of the product use to the customer’s unique needs and usage situation. This distinction is based on how value is co-created and appropriated by the servitized unit and the customer, which is at the core of explaining performance (Kastalli and van Looy, 2013). Our main contention is that BAS and ADS require different capabilities and have different associations with performance, resulting in naturally occurring servitization trajectories. We test our model with data from a large survey of manufacturers from different countries and sectors.

4

2. Theoretical background 2.1. A value-based perspective of servitization strategies Central to understanding servitization strategies and their performance outcomes are the processes for co-creating value with the customer through services (Kindström, 2010; Kastalli and van Looy, 2013). We employ Gronroos’s (2011) service logic to frame these processes. Under this logic, a non-servitized manufacturer provides customers with value potential only, embedded in the manufactured products. The customer creates value in use by using the products without interacting with the manufacturer. The products act as input resources into the customer’s processes of value creation and the manufacturer has no direct control over how they unfold (Gronroos, 2011). In contrast, a servitized manufacturer not only provides value potential, but also co-creates value with the customer by adding services to products. These services affect the way by which customers create value and the manufacturer can influence the customers’ processes of value creation directly (through interactions). We therefore argue that servitization strategies are defined to a great extent by the types of services offered by the manufacturer (Cusumano et al., 2008; Eggert et al., 2014), most notably by the associated nature of value co-creation that takes place with the customer (Smith et al., 2014). Specifically, we distinguish between Basic Services (BAS) and Advanced Services (ADS) (Gebauer et al., 2005; Mathieu, 2001a). BAS aim to install and maintain basic product functionality in an efficient and effective manner for the customer. Examples are product installation, provision of spare parts, maintenance and repair. BAS have limited association with how customers create

5

value because they involve limited customer interaction and the value that is co-created does not go beyond that associated with effortless availability of the basic functions of the product (i.e., the value potential embedded in the product). BAS broadly correspond to what prior research has labeled product-oriented services (Gebauer et al., 2005; Mathieu, 2001a; Neely, 2008; Smith et al., 2014). ADS relate to working closely with customers to co-create value that goes beyond basic product functionality, involving the customer’s actions in relation to the product and the adaptation of the product use to the customer’s unique needs, usage situation and behaviors (Mathieu, 2001a; Smith et al., 2014). Examples are training in using the product, product upgrades, consulting and product rental. ADS broadly corresponds to what Gebauer et al. (2005) call customer support services, including use-oriented and resultoriented services (Mathieu, 2001a; Neely, 2008; Tukker, 2004; Smith et al., 2014). ADS are a key distinguishing feature of the concept of integrated solutions, consisting of bundles of products, BAS and ADS that form high-value, complex and result-oriented offerings that address a specific customer’s need (Brax and Jonsson, 2009; Davies, 2004). Table 1 summarizes the differences between BAS and ADS across relevant dimensions, based on the literature on servitization and product-service systems. [Table 1] In order to understand the capability antecedents and performance outcomes of servitization, we adopt Kastalli and van Looy’s (2013) framework to analyze value creation and appropriation processes associated with BAS and ADS. In this framework, service offerings affect financial performance through supply-side effects (capabilities) and demand-side effects (product substitution and customer perceived value). We focus on two 6

financial performance dimensions generally associated with servitization: sales and profitability. On the supply side, servitization may increase costs by requiring investment in service-specific capabilities. On the demand side, servitization may increase revenue by complementarity effects, i.e., customers reap greater value from buying a bundle of products and services from a single supplier than from buying separate products and services from different providers. Customers benefit from bundles due to increased interoperability, reduction in procurement costs, and exploiting supplier manufacturing competencies, e.g., through product maintenance (Kastalli and van Looy, 2013). This added value may increase the premium price of the product-service bundle (Eisenmann et al., 2011). However, servitization may also reduce revenue through substitution effects, i.e., product life cycles increase due to better services, so customers postpone the purchase of new (replacement) products (Kastalli and van Looy, 2013). 2.2. Theoretical model The theoretical model is depicted in Figure 1. We posit that BAS requires manufacturingbased capabilities, while both BAS and ADS require service-specific capabilities. We also hypothesize that offering BAS is a necessary condition for offering ADS. Moreover, we argue that BAS has a neutral or negative impact on financial performance, while ADS has a positive impact. [Figure 1] Based on these arguments, we contend that there are natural trajectories to develop servitization over time, based on a balanced growth of BAS and ADS provision in tandem

7

(i.e., BAS increase first, followed by ADS). This is shown in Figure 2. Although units can exhibit different servitization strategies (mix of BAS and ADS), we argue that these strategies will naturally occur below and close to the solid diagonal of Figure 2. That is, ADS builds on BAS, with BAS acting as a platform to sell ADS. The corresponding hypotheses are developed next. [Figure 2] 3. Hypotheses 3.1. Capabilities for servitization For the purpose of this study, we adopt the definition of capability as the capacity of the organization to “perform a particular activity in a reliable and at least minimally satisfactory manner” (Helfat and Winter, 2011, p. 1244). Consistent with our view of services as processes (Sampson and Froehle, 2006), we employ a definition of servitization capability focused on a manufacturer’s ability to carry out the management (design and delivery) of the service processes it offers, repeatedly and reliably (Winter, 2003). Different types of capabilities for servitization have been discussed in the literature (e.g., Oliva and Kallenberg, 2003; Fang et al., 2008; Martinez et al., 2010; Storbacka, 2011; Ulaga and Reinartz, 2011; Kastalli and van Looy, 2013). We argue that BAS and ADS, because of their different nature (Table 1), are related to different domains of knowledge and skills, namely, those related to managing manufacturing processes (product-centered and noninteractive) and service processes (customer-centered and interactive) (Nambisan, 2001). Hence, they require different sets of capabilities.

8

BAS require manufacturing-based capabilities, namely, special knowledge about product design, its technology and product/process engineering (Oliva and Kallenberg, 2003; Ulaga and Reinartz, 2011). For example, maintenance services require knowledge about product service requirements over its life cycle, and fabricating spare parts requires specialized production technologies. Technical expertise and credibility are key for customers to buy product-based services (Ulaga and Reinartz, 2011). On the other hand, ADS are not product-centric, and thus should not depend directly on manufacturing capability development. We put forward the following hypotheses: H1a. Manufacturing capabilities are positively associated with BAS provision. H1b. Manufacturing capabilities are not associated with ADS provision. Service-specific capabilities are required for the design and delivery of both BAS and ADS. The servitized manufacturer requires the ability to design services and products jointly (e.g., product design for serviceability; Kindström and Kowalkowski, 2009; Ulaga and Reinartz, 2011). It also requires skills to design and manage service processes, which, compared to manufacturing processes, are more people-intensive and require more frequent and complex interactions with customers (Kindström and Kowalkowski, 2009). Service provision should also be backed by a well-functioning service organization (Oliva and Kallenberg, 2003), which may involve the creation of a separate organization to handle service provision, building a dedicated sales force and service technicians, changing the incentive and control systems in order to foster integrated product-service performance and developing service-centered employee skills and values (e.g., customer- and service-based values/orientation, autonomy, customization, flexibility, quick response, etc.) (Baines and Lightfoot, 2014; Gebauer et al, 2005; Oliva and Kallenberg, 2003). 9

We posit the following hypotheses: H2a. Service capabilities are positively associated with BAS provision. H2b. Service capabilities are positively associated with ADS provision. 3.2. BAS as necessary condition for ADS We posit that BAS is a necessary condition (platform) for providing ADS. Both demandside and supply-side arguments support this hypothesis. From the demand side, recall that the focus of our study is on manufacturers that may or may not offer services alongside products. Their customer purchases a product that often needs support through the life cycle. Therefore, the customer likely expects (and even demands) BAS (Oliva and Kallenberg, 2003; Martinez et al., 2010). If the manufacturer does not offer BAS, and since these services are often not complex, the customer may choose to service the product in-house, use an independent services provider or even take the whole business (product and product-centric service) to another manufacturer (Kastalli and van Looy, 2013). This may put the manufacturer into a virtually untenable market position to offer ADS (Kindström and Kowalkowski, 2009). As discussed earlier, the customer may obtain significantly greater value from acquiring a bundle of products and services (BAS and ADS) from a single provider than from using separate product and service providers. Product-service procurement may also have dynamic effects. By offering BAS, the manufacturer can build stronger relationships and trust with the customer (Kindström and Kowalkowski, 2009). As the customer benefits from the enhanced value proposition from the product-BAS bundle, and increases confidence in the manufacturer’s ability to provide 10

quality services, it becomes more willing to contract higher-risk ADS (Gebauer et al., 2005). Thus, the manufacturer’s entry into the service market requires offering BAS first, and this may facilitate offering ADS later. On the supply side, the key argument is that offering ADS requires a deep understanding of how customers create value by using products (Brax, 2005; Kindström and Kowalkowski, 2009). A number of contextual factors such as the customer’s strategic goals, equipment knowledge, and the equipment use physical environment may affect customer value creation through product utilization (Smith et al., 2014). Since these factors may be heterogeneous and customer-specific, understanding customer value creation requires interactions with the customer during product usage encounters (Payne et al., 2008; Kindström and Kowalkowski, 2009). We argue that such understanding is mainly acquired by providing BAS. As an entry vehicle into the service market, BAS provide a natural opportunity for manufacturers to map the installed product base and learn about customers and product utilization (Ulaga and Reinartz, 2011). For example, by gathering historical product failure data, manufacturers can estimate risk and develop better pricing policies for product rental services; by observing product usage patterns, they may co-design offerings that are more tailored to customer needs (Porter and Heppelmann, 2014). Therefore, both demand and supply side arguments suggest that ADS offerings must follow a corresponding or greater level of BAS offerings. These arguments lead to the following hypothesis: H3. The offering of BAS is a necessary condition for the offering of ADS.

11

3.3. Impact of BAS and ADS on financial performance We argue that performance impacts of BAS and ADS are different due to alternative demand and supply-side effects. BAS focus on ensuring basic product functionality. The perceived value of a product-BAS bundle is only modestly higher than the value of separate products and services available from different providers. BAS offerings do not provide significant differentiation (Gebauer et al., 2005), and must compete against specialist service providers or even customer in-house units (e.g., internal maintenance department). In some cases, customers expect these services to be provided freely, and a product-centric manufacturer may be tempted to give them away or at below cost to promote its own products (Ulaga and Reinartz, 2011; Suarez et al., 2013). Thus, service sales margins will be low or even negative (and potentially below product sales margins) (Gebauer et al., 2005; Suarez et al., 2013). As such, BAS may contribute only moderately to service sales revenues. Moreover, BAS aim to improve product functionality, which prolongs product life and results in product substitution effects. Therefore, product sales may decrease rather than increase with BAS. Overall, BAS may not have a significant impact on sales and margins. Although the customer receives an enhanced value proposition, the provider is not able to appropriate that value. On the supply side, providing BAS requires investments in manufacturing and service capabilities. Still, cost increases can be minimized because BAS-related capabilities may not be radically different from product-centered capabilities. For example, because BAS involve quasi-manufacturing processes (Table 1), they can benefit from manufacturing-based process management and labour skills (Bowen et al., 1989; Kastalli 12

and van Looy, 2013). Besides, BAS may not require high levels of service capabilities, for example, they may exclude the creation of a separate service unit (Oliva and Kallenberg, 2003). Overall, BAS provision is expected to lead to a moderate increase in costs. These arguments lead to the following hypotheses: H4a. The provision of BAS has no significant impact on sales. H4b. The provision of BAS has a negative impact on profitability. ADS focus on the co-creation of value in use for customers beyond that which is embedded in basic product operation. The perceived value of a product-ADS bundle, especially if configuring a solution, may be significantly higher than the value of separate products and services (Davies, 2004). ADS provide significant differentiation, creating strong customer lock-in and loyalty effects (Ulaga and Reinartz, 2011). Accordingly, premium prices, and thus service sales margins are higher. Besides, ADS do not specifically target product functionality, and thus should not cause product substitution effects. Overall, ADS provision may have a positive impact on sales and margins. On the supply side, ADS require significant investments in service-specific capabilities. For example, they may require a separate service unit (Oliva and Kallenberg, 2003) and are more knowledge- and people-intensive (Schmenner, 2004; Von Nordenflycht, 2010). Thus, ADS provision costs are higher than BAS costs. However, because these are highly differentiated services, providers should still be able to appropriate a significant part of the enhanced customer value proposition (through premium prices for ADS), resulting in net sales margin increases. Thus, we put forward the following hypotheses:

13

H5a. The provision of ADS has a positive impact on sales. H5b. The provision of ADS has a positive impact on profitability. 3.4. Servitization trajectories Based on the previous hypotheses (Figure 1), we expect to find a naturally occurring servitization trajectory, from a baseline manufacturing business model towards increased levels of servitization maturity. By servitization maturity we mean the extent to which the business model is reliant on service provision versus products (Kowalkowski et al., 2015). In this trajectory, the provider gradually introduces BAS to learn about how customers create value through product usage in the customer’s specific context, and follows with ADS afterwards. BAS and ADS work in tandem: the provider starts the service relationship with a customer with BAS and consolidates its position by offering ADS. Over time, this pattern (BAS, followed by ADS) is extended to a larger number of customers, resulting in increased

servitization

maturity

(Figure

2).

BAS

work

for

service

market

penetration/breadth (offering services to additional customers), followed by ADS for developing market depth (offering higher levels of service (ADS) per customer) in tandem. The trajectory is supported by the following arguments. BAS are offered first because they draw on existing manufacturing capabilities (H1). They facilitate a cultural change towards service (Gebauer et al., 2005) and allow time to improve service capabilities for broader service offerings (H2). ADS are offered later; they require BAS not only for market entry but also to learn how individual customers create value through product usage (H3). The trajectory end goal is to offer as high a level of ADS as possible

14

(using BAS as a platform), since it is ADS and not BAS that associate positively with performance (H4-H5). Prior literature has suggested that servitization trajectories involve providing BAS first to a high extent across customers, and only then follow with ADS. That is, BAS and ADS offerings correspond to distinct and sequential stages along a transition path (Oliva and Kallenberg, 2003; Gebauer et al., 2005; Martinez et al., 2010; Eggert et al., 2014). While we agree that BAS are necessary for ADS at the level of individual customers, we argue that a trajectory in which a provider offers BAS significantly across customers but not ADS (i.e., market breadth without market depth) is less sustainable and therefore less likely to occur. This is because BAS do not provide sufficient returns to support capability investments, or internal momentum to maintain motivation and credibility towards servitization (Gebauer et al., 2005). Managers will only aim to improve service offerings if they consider this to be strategically relevant for the business (Suarez et al., 2013); this is more likely to occur through ADS than BAS. Because ADS are customized, economies of scale are less important, so that it is viable to start offering ADS (service depth) to individual customers after BAS have been introduced to those customers, without necessarily having an extensive breadth of BAS offerings across customers. Therefore, we propose a trajectory in which BAS and ADS and developed in parallel (combining market breadth and depth) (Kowalkowski et al., 2015), rather than strictly sequentially (BAS first with high breadth followed by ADS). We operationalize this trajectory by the following hypotheses: H6. Higher levels of offering of ADS and BAS (with BAS as necessary condition for ADS) correspond to higher levels of servitization maturity. 15

H7. ADS relates positively with BAS in manufacturing units operating at both low and high levels of servitization maturity. H6 intends to ascertain whether manufacturers with higher levels of BAS and ADS offerings indeed display a business model with higher reliance on services. H7 intends to ascertain whether manufacturers exhibit a balanced offering of BAS and ADS at different stages of the trajectory, ruling out an alternative trajectory based on a much higher offering of BAS than ADS at initial maturity levels.

4. Data We used data from the Sixth International Manufacturing Strategy Survey (IMSS-VI). These data enabled to test hypotheses in a broad and international manufacturing set, while controlling for industry and country covariates. Moreover, data were collected at the business unit level, which was consistent with our model of servitization practices, capabilities and performance. The following information appeared in previous studies that used IMSS data (e.g., da Silveira (2011), Tian et al. (2012), Hong et al. (2014), Golini and Kalchschmidt (2015), Longoni and Cagliano (2015)) or was available from documents distributed by the survey network (IMSS, 2015). IMSS is a global periodic survey of manufacturers of fabricated metal products, machinery and equipment (ISIC 25-30). It is focused on operating strategies, practices, and performance of manufacturing units. An international network of operations strategy scholars carries out the survey.

16

IMSS-VI was carried out in 2013-2014 in 22 countries. It included 931 valid responses from Belgium (29), Brazil (31), Canada (30), China (128), Denmark (39), Finland (34), Germany (15), Hungary (57), India (91), Italy (48), Japan (82), Malaysia (14), The Netherlands (49), Norway (26), Portugal (34), Romania (40), Slovenia (17), Spain (29), Sweden (32), Switzerland (30), Taiwan (28) and USA (48). The original instrument was developed in English and tested with pilot respondents. The English version was used in 11 countries and versions in other languages (mostly based on double parallel translations) were used in the remaining 11 countries. The Director of Manufacturing /Operations or equivalent in the unit responded to the questionnaire. Respondents were initially contacted by email or phone. The response rate was 13%: 2586 of the initial 7167 contacts agreed to participate; they returned 1003 questionnaires, of which 931 were considered valid (i.e., had no more than 60% missing fields). Respondents had on average 13.2 years of experience in the company (N = 894), of which 11.5 had been in operations or manufacturing (N = 866). In all but two countries, researchers checked for non-respondent and late-respondent bias by testing differences in market performance (sales and return on sales) between respondents and non/laterespondents; no significant differences were found. Survey administrators also carried out several data reliability tests; when dubious responses were found, they contacted respondents for clarification, or in a few cases dropped questionnaires that did not appear reliable. As indicated by da Silveira (2011) and Longoni and Cagliano (2015), the IMSS survey incorporates several features to minimize common method bias (Podsakoff et al., 2003). The instrument is quite long and suggests no specific links between strategies, 17

practices, and performance. Respondents are advised that responses are treated anonymously and with confidentiality. Finally, questions and scales are clearly presented. We know of three previous studies that used IMSS data to validate models of servitization or service orientation. Tian et al. (2012) used a previous dataset (IMSS-V) in a study that supported the moderating role of "service capability" on the relationship between "service delivery" and "business performance". Different from our model, they merged servitization items into a single construct that did not differentiate between basic and advanced services. Based on an older dataset (IMSS-IV), Hong et al. (2014) suggested that "strategic customer service orientation" (a different construct composed by service priority items) had a positive impact on lean practices and operational performance. A recent paper by Kuula and Kauppi (2015) used IMSS-VI data to test the impact of "servitization" on "product development" and performance. However, their servitization scale is based on capability items rather than service offerings, whereas our model clearly differentiates between these two concepts. 5. Measures 5.1. Capabilities The capabilities scales were based on theory (section 3.1) and the survey development (IMSS, 2015). Manufacturing capabilities for servitization refer to special knowledge about product and process engineering (Oliva and Kallenberg, 2003; Ulaga and Reinartz, 2011). The manufacturing capability scale (MANCAP) reflected the manufacturer’s ability to operate with complex products and processes (Lin and Uhler, 2002; Mihm et al., 2010). 18

This complexity included three aspects, which influenced the scale development (IMSS, 2015): (i) the number of unique product configurations that manufacturing must produce in the absence of modular designs (Ericsson and Erixon, 1999; Lin and Uhler, 2002), (ii) the number of different parts or materials in process, which lead to intricate bills of materials (Khan et al., 2012), and (iii) the number of processing tasks to be completed for different product configurations (Leachman and Carmon, 1992). MANCAP included three items corresponding to these issues. The question was, “How would you describe the complexity of the dominant activity?” Responses were given on five-point scales with the following anchors: [MCP1] (1) “Modular product design”, (5) “Integrated product design”; [MCP2] (1) “Very few parts/materials, one-line bill of material”, (5) “Many parts/materials, complex bill of material”; [MCP3] “(1) “Very few steps/operations required”, (5) “Many steps/operations required”. To assess the content validity of the MANCAP scale, we carried out bivariate correlation analyses between MCP1, MCP2, MCP3 and alternative indicators of manufacturing capability available from the survey. The indicators measured the current level of implementation of “continuous improvement programs”, “workers flexibility” programs, “advanced processes” such as “high precision technologies”, “factory of the future” developments, and “quality improvement and control” programs. Thus, indicators covered a range of “soft” and “hard” programs aimed at increasing the manufacturing capability to produce and service products or components. All three MANCAP indicators correlated positively and significantly (p < .01) with all five alternative indicators. The results indicated that MANCAP was a valid operationalization of manufacturing capability at the plant.

19

Consistent with our theoretical background, the service capabilities scale (SERVCAP) focuses on broad capabilities that support the design and delivery of service processes (which are customer-centered and interactive). Respondents were asked to indicate the current level of implementation of these capabilities. Responses were on a fivepoint scale with anchors 1 (“None”) and 5 (“High”). They were measured by three items. SCP1 [“Expanding the service offering to your customers (e.g., by investing in new service development”] reflects the need to develop capabilities to design new services aligned with evolving customer needs (Kindström and Kowalkowski, 2009; Kowalkowski et al., 2015). SCP2 [“Designing products so that the after sales service is easier to manage/offer (e.g., design for maintenance)”] reflects the need to develop capabilities to design services and products jointly (Ulaga and Reinartz, 2011), incorporating increased customer-induced variability. SCP3 [“Developing the skills needed to improve the service offering”], reflects the need to develop broad organizational skills (including workforce) to provide better service processes to customers over time (Ulaga and Reinartz, 2011). 5.2. Service offerings Consistent with our research model, we measured the intensity (or volume) of the provision of BAS and ADS. We drew on an IMSS question asking respondents to rate from 1 (“None”) to 5 (“High”) the extent that a list of services were “offered alongside with the products by the business unit”. The question comprised a list of eight services typically associated with servitization strategies in the literature (Gebauer et al., 2005; IMSS, 2015; Neely, 2008; Smith et al., 2014). We classified each of these services as Basic or Advanced, following closely our adopted definitions for BAS and ADS (see section 2.1). The variable BAS comprised services whose objective was to set-up and maintain basic 20

product functionality. They corresponded broadly to Gebauer et al.’s (2005) category of product-oriented services, and included three items: “maintenance and repair of products” (BAS1), “installation/implementation” (BAS2), and provision of “spare parts/consumables” (BAS3). ADS comprised services related to co-creating value that goes beyond basic product functionality, involving the customer’s unique needs and product usage situation. They corresponded broadly to Gebauer et al.’s (2005) category of customer support services, and included five items: “rental/lease of products (with responsibility for maintenance, repair and operation)” (ADS1) (in this case, the provider took on a number of activities that were usually internal to the customer), “product upgrades (software, product modifications)” (ADS2) (in the sense that these upgrades needed to match customized customer requirements arising throughout product usage), “helpdesk/customer support center” (ADS3), “training in using the products” (ADS4) and “consultancy services” (ADS5). 5.2. Financial performance We operationalized business unit financial performance by sales (SALES) and profitability. Profitability was assessed by the commonly used measure of return on sales, defined as net profits (sales revenue minus total costs, before interests and taxes) as a percentage of sales (Suarez et al. 2013). Respondents were asked, “Please indicate your Sales and Return On Sales of the business unit in 2012”. Sales responses were on a five-point scale with values 1 (< 10 Million €), 2 (10-50 Million €), 3 (50-100 Million €), 4 (100-500 Million €), and 5 (> 500 Million €). ROS responses were also on a five-point scale with values 1 (< 0%), 2 (05%), 3 (5-10%), 4 (10-20%), and 5 (> 20%).

21

According to the IMSS administration (IMSS, 2015), these indicators were adapted from Venkatraman and Ramanujan (1986). We used correlations between our variables and corresponding “growth” indicators to assess the reliability of responses. Respondents had been asked how much each indicator had changed over three years. They used a five-point scale with endpoints 1 (“Much lower”) and 5 (“Much higher”). The correlations between SALES and sales growth (r = .186, p < .001, n = 854) and between ROS and ROS growth (r = .317, p < .001, n = 830) were both positive and significant. 5.3. Control variables We developed three sets of control variables. They had been used as covariates in previous servitization studies, e.g., Lay et al. (2010). The first was unit size. Larger units may not only sell more, but also have more resources to innovate (Gorodnichenko et al., 2010) and thus develop new capabilities and services. We measured SIZE by the number of employees in the business unit. Industry sector could also have a significant influence on endogenous and exogenous variables of the model. Complex products such as electrical (ISIC 27) and large transportation equipment (ISIC 30) may require more customer-oriented capabilities and services than metal products such as parts and components (ISIC 25) (Zhang and Zhang, 2014). In addition, the stage of an industry’s technology lifecycle may affect the types of services offered by manufacturers (Cusumano et al, 2015). Thus, we controlled for ISIC sector. Since the survey included respondents from ISIC codes 25 to 30, we created five binary variables (ISIC25, ISIC26, ISIC27, ISIC28, ISIC29) with value one if a respondent was classified in that sector, and zero otherwise.

22

Finally, country endowments can leverage capabilities, services, and performance in manufacturing. Plants operating in more developed countries may have greater access to organizational and technological resources, which in turn support innovation and performance (Neill et al., 2014). We controlled for national endowments by GDP per capita (current prices) in 2012 (IMSS-VI baseline year) of the plant’s country of operation (GDPCAP). We obtained data from the IMF’s World Economic Outloook Database (IMF, 2015). The variables SIZE and GDPCAP were transformed by the natural log (ln) before the main analyses to improve normality. 5.4. Descriptive statistics We inspected all variables for outliers and anomalous statistics. They had a relatively low percentage of missing values (between 0% and 8%). Thus, listwise deletion of cases with missing data was acceptable if they were missing completely at random (MCAR) (Fichman and Cummings, 2003; Honaker and King, 2010). To test the MCAR hypothesis, we performed Little’s (1988) test with the Missing Value Analysis procedure and expectationmaximization (EM) estimation in SPSS®. The χ2 statistic was non-significant (p = .678), so data could be considered MCAR. Thus, tests of hypotheses H1-H5 included only cases with complete responses (N = 763). We present descriptive statistics in Table 2. SERVCAP, MANCAP, and BAS item means were near the centre of the scale. Most ADS item means were in the lower half, which was consistent with the theoretical model. Mean ROS was near centre, but mean SALES was in the lower half of the scale. This was likely because the sample included

23

many respondents from small and medium-sized units, and from medium and medium-low income economies. ISIC dummies were somewhat uneven with 55% of respondents belonging to sectors 25 (fabricated metal products) or 28 (general machinery and equipment). Unit sizes varied significantly (Q1 = 100, Q2 = 270, Q3 = 950) as the sample was well represented by small, medium-sized, and large units. [Table 2] 6. Analysis and Results We tested H1-H5 through partial least squares structural equation modeling (PLS-SEM). The path model (Figure 1) was implemented with the plspm package (Sanchez et al., 2015) in R version 3.2.2 (R Core Team, 2015). The procedures were based on Sanchez’s (2013) manual and recommendations in Hair et al. (2014). We used PLS-SEM rather than covariance-based SEM (CB-SEM) for three reasons. First, histograms and Q-Q plots indicated that key variables departed from normality, which is an assumption in CB-SEM but not in PLS-SEM (Hair et al., 2011; Sanchez, 2013; Hair et al., 2014). Second, the model included many single-item control variables, which PLSSEM (but not CB-SEM) would process without identification problems (Hair et al., 2011; Hair et al., 2014). The third reason was theoretical. As explained by Hair et al. (2014), PLSSEM is focused more on prediction than explanation, “… which makes PLS-SEM particularly useful for studies on the sources of competitive advantage and success driver studies” (p. 78). As shown in Table 2, the analysis included 23 observed and four latent variables (LVs). All LVs were built as reflective scales. Following MacKenzie et al. (2011), we 24

considered our latent constructs (the overall focus on services or capabilities) as causes rather than consequences of investments in specific service offerings or skills in the unit. Arguments for the plspm() function were based on recommendations in Hair et al. (2014). The weighting scheme was “path”, the scaled option was “TRUE” so data were zstandardized before estimating latent variable scores, tolerance (tol) was 1e-5, the maximum number of iterations (maxiter) was 300, and bootstrap validation (boot.val) was “TRUE” with 10000 resamples (br). 6.1. Measurement model We used measurement model estimates (Table 3) to assess the reliability, convergent validity, and discriminant validity of scales. All the following tests and threshold values were based on recommendations in Sanchez (2013) and Hair et al. (2014). Cronbach’s alphas were near to, or above .7, while composite reliability estimates were all above .7, supporting the reliability of estimates. Two indicator loadings (ADS1, MCP1) were slightly below .7; they were not removed because reliability and convergent validity estimates of the ADS and MANCAP scales were above the minimum thresholds. All four AVE estimates were above .5, supporting convergent validity. Correlations with the four latent variables (Table 4) were all below the square roots of their respective AVEs, which supported discriminant validity. Unidimensionality was supported as the first eigenvalues of the scales were much higher than one, while the second eigenvalues were below one.

25

Table 3 also presents the item weights used to estimate latent variable scores. The scores were used to estimate Pearson correlations (Table 4) and the SEM path estimates (next section). [Table 3] [Table 4] 6.2. Structural model The quality of a PLS-SEM is assessed by predictive criteria rather than measures of fit (Hair et al., 2014). Thus, we assessed the quality of the structural model by four conditions (Wetzels et al., 2009; Sanchez, 2013; Hair et al., 2014): (i) significance of path estimates, (ii) coefficients of determination of endogenous variables, (iii) effect sizes of exogenous variables, and (iv) model goodness-of-fit (GoF). 2 Path estimates (excluding control variables) and partial effect sizes (𝑓𝑥𝑦 ) with the

four endogenous variables (Hair et al., 2014) are presented in Table 5. The significance of path estimates was established by confidence intervals at ±1.96 times standard errors obtained from bootstrapping (α = 0.05) (Hair et al., 2014). Seven of the nine path estimates were significant. Coefficients of determination (𝑅𝑦2 ) of the four endogenous variables were .280 (BAS), .537 (ADS), .410 (SALES) and .042 (ROS). Therefore, the model had good predictive capabilities for all endogenous variables except ROS. Six of the nine 2 𝑓𝑥𝑦 exceeded the minimum threshold suggested by Cohen (1988) and reported by Wetzels

et al. (2009). Together, path estimates and partial effect sizes supported all hypotheses except H4b (the BAS -> ROS estimate was significant and negative, but the effect size was below the minimum threshold in Cohen (1988)). 26

[Table 5] SALES was the only of four endogenous variables with significant associations with control variables. It related positively with ln-SIZE and ln-GDPCAP, and negatively with ISIC25, ISIC26, and ISIC28 (p < .05). Thus, (i) units with more employees and located in countries with greater GDP per capita had greater revenue and (ii) units from ISIC sectors 25, 26, and 28 had lower revenue. GoF is calculated by the geometric mean of (i) measurement model commonalities (i.e., the square of indicator loadings – see Table 3) and (ii) 𝑅𝑦2 estimates (Wetzels et al., 2009; Sanchez, 2013). The model GoF was .449, above the .36 “large” threshold in Wetzels et al. (2009). Therefore, the model appeared to report overall large effects of exogenous on endogenous variables. 6.3. Robustness test of H3 Whereas most hypotheses in our model predicted co-variation between endogenous and exogenous variables, H3 was more specific. Based on the literature on servitization, we predicted that BAS were a “necessary condition” to ADS. As indicated by Dul et al. (2010), such hypotheses are categorically different from co-variation hypotheses and require additional testing approaches beyond regression. Latent variables including BAS and ADS were estimated by weighted averages of standardized items. Thus, even though observed indicators were based on ordinal scales, their estimates were treated as continuous variables. Dul et al. (2010) suggested testing “necessary condition” hypotheses by continuous variables based on scatterplots with the condition variable in the x-axis and the outcome variable in the y-axis. 27

Figure 3 presents the scatterplot of BAS and ADS. It shows not only a positive correlation between BAS and ADS (as found earlier), but also a clear “ceiling” (Dul et al., 2010) for levels of ADS at given levels of BAS. This result provided further support to H3 that offering basic services was a “necessary condition” to offering advanced services by manufacturers. [Figure 3] 6.4. Cluster analysis We tested H6 and H7 (servitization trajectories) by identifying groups of manufacturing units with distinct patterns of offering of BAS and ADS, and by characterizing the corresponding level of servitization maturity (as an indication of advancement along a servitization trajectory). We employed cluster analysis with BAS and ADS as the clustering variables, using summated scales (the sum of scores of individual items divided by the number of items in the scale). Servitization maturity was assessed by three variables in the survey: i) the percentage of sales turnover invested in strategic initiatives including “sustainability, globalization and servitization” (services investment); ii) the share of sales turnover that was based on services as opposed to parts and components, and assembled products (services sales); iii) the average (five-point) importance given to two service priorities (product assistance/support, customer service) as order-winners from major customers (services priority). We performed the analysis with the complete dataset (n = 931), using IBM® SPSS® Statistics 22 (IBM Corp., 2013) in two stages, following guidelines in Timm (2002) and Hair et al. (2010). We initially carried out hierarchical analyses with Squared Euclidean distances and three alternative methods: Ward’s, between-groups linkage, and within28

groups linkage. We had two objectives at that stage. First, we reviewed distance coefficients and cluster sizes to determine the best number of k groups. In all cases, two groups seemed the best solution. Second, we estimated cluster averages from each solution to use as initial centers for k-means (non-hierarchical) analyses in the second stage. In the second stage, we performed three k-means analyses, each using centers from one hierarchical method. We calculated cluster stability coefficients between hierarchical and k-means solutions. Since coefficients were 85.19% (within-groups linkage), 94.28% (Ward’s) and 95.51% (between groups linkage), we used the third solution in subsequent analyses. Following Hair et al. (2010), we further validated that solution by re-sorting observations based on an unrelated variable (SIZE) and carrying out a new k-means analysis. This solution had 95.70% stability. Since BAS and ADS were higher in cluster 2 than in cluster 1, we called cluster 1 “Low Services” and cluster 2 “High Services”. Table 6 presents the ANOVA between clusters. The variables did not pass Levene’s test of homogeneity of variances and, perhaps more importantly (see Zimmerman, 2004) size differences between clusters appeared to be significant. Based on these outcomes and the recommendation in Zimmerman (2004), we estimated robust Welch statistics rather than ANOVA F. The High Services cluster had significantly higher levels of all variables than the Low Services cluster. Thus, the High Services cluster appears to correspond to a higher level of servitization maturity (i.e., a more advanced stage of servitization) than the Low Services cluster. Besides, BAS was higher than ADS in both clusters. These results support H6 that manufacturers with higher levels of BAS and ADS offerings (with BAS as a necessary condition for ADS) indeed display higher servitization maturity.

29

Even though clusters corresponded to significantly different stages of servitization maturity, the mean level of adoption of BAS and ADS were balanced in both clusters (1.98 - 1.76, 3.96 - 3.20). There was also a significant correlation between these two variables in each cluster (r1 = .243, p < .001; r2 = .214, p < .001). These results support H7 that manufacturers exhibit a balanced offering of BAS and ADS at different stages of maturity, ruling out an alternative trajectory based on a much higher offering of BAS than ADS at initial maturity levels. [Table 6] 7. Discussion Our findings provide new insights into the servitization paradox and servitization trajectories. They suggest that the decline in profits often reported in initial stages of servitization may be caused by the fact that at such stage of servitization - characterized by low levels of BAS and ADS - the rise in costs is not made up by sufficient returns, since BAS have a neutral or detrimental impact on profits and the level of ADS provision is low. It is only when units offer significant levels of ADS that they achieve increased performance. Thus, the findings point to naturally occurring servitization trajectories based on a balanced adoption of BAS and ADS, using BAS as a platform, rather than BAS first to a high extent, followed by ADS. BAS work for service market penetration (offering services to additional customers), followed by ADS for developing market depth (offering higher levels of service (ADS) per customer) in tandem. Figure 3 supports this argument by showing that most manufacturing units are located close to the BAS=ADS diagonal and there are few units in the upper-left and lower-right quadrants.

30

It is useful to compare our trajectory findings with other studies. Our theoretical model focuses on firm-level strategic factors, namely the manufacturer’s internal capabilities and competitive strategy, as the key explanatory factors determining servitization trajectories. Consistent with most studies on servitization paths, we consider as the baseline model a pure manufacturer and frame these firm-level factors as drivers of increasing levels of servitization maturity. Some studies (e.g., Finne et al., 2013; Turunen and Finne, 2014; Cusumano et al., 2015) conceptually propose that changes in industry contextual variables, such as technology maturity and regulatory environment, may lead to reverse servitization, by which at some point a manufacturer may decrease its level of servitization maturity. Since our study controls for industry, it partially accounts for these potential effects. However, the investigation of fine-grained back and forth movements along a trajectory caused by industry level variables, with possible hysteresis, falls outside the scope of a cross-sectional study such as ours. We submit that the firm-level drivers that underlie our proposed trajectory apply generally, although they may be overlaid by the influence of broader industry-level variables that have longer timescales. Our proposed trajectory contrasts with prior studies suggesting that manufacturers need to develop and consolidate the portfolio of BAS first, followed by a systematic development of ADS; that is, BAS and ADS offerings correspond to distinct and sequential stages along a transition path (Oliva and Kallenberg, 2003; Gebauer et al, 2005; Martinez et al, 2010; Eggert et al, 2014; Kowalkowski et al., 2015). While we concur that BAS are necessary for ADS at the level of individual customers, we argue that a trajectory in which a provider offers BAS significantly across customers but not ADS (i.e., market breadth without market depth) is less sustainable and therefore less likely to occur. Extant studies

31

did not include measures for capabilities and did not examine the link between the adoption of different servitization strategies and servitization maturity. By integrating capability antecedents, performance outcomes and servitization maturity measures in a single study, we were able to provide strong empirical support for our assertions on how BAS and ADS enact the servitization paradox and servitization trajectories. Our findings are consistent with Kowalkowski et al.’s (2015) view of servitization trajectories as building on product-BAS bundles by adding increasingly complex services, but

without

ever

discarding

the

supply

of

products

and

BAS.

These authors conceptually discuss three possible trajectories. The availability provider trajectory involves adding use-oriented services to a product-BAS bundle. The performance provider trajectory involves adding performance-oriented services to either a product-BAS bundle or an availability provider bundle. Since in our study we group use-oriented and performance-oriented services under ADS, we are not able to differentiate between the availability provider and performance provider paths, but both of these paths are consistent with moving along the diagonal of our proposed trajectory towards solution-type offerings (products, BAS and ADS) (Figure 2). The third path – industrializer – involves extending the offer of BAS to more customers, on top of the availability provider bundle. This is consistent with our proposition of a balanced development of BAS and ADS along the servitization trajectory. Our findings raise questions about the role played by BAS in servitization strategies. We identify two roles for BAS. First, BAS are expected to increase the value offered to customers, despite the provider having difficulties in appropriating part of this value. In some cases, BAS are offered due to market forces, including coercive pressures from 32

customers (Martinez et al., 2010). Thus, BAS can play a defensive role for the product business, working as a market qualifier and avoiding customer defections to competitors (Eggert et al., 2014). The adoption of BAS in isolation corresponds to a product-centric business model that considers services as a “necessary evil” (Ulaga and Reinartz, 2011), not radically different from a pure manufacturing model. It may be less sustainable in the long-term. Second, although BAS are not a source of revenues and profits in themselves, they are necessary to penetrate the service market and support profitable business models, based on the provision of ADS and synergies between BAS and ADS. Thus BAS can also play an offensive role for the service market, working as platform for ADS provision and service-based competitive strategies (Baines et al., 2010).

8. Conclusions The study is one of the first to provide a theoretical articulation and empirical test of an integrated, fine-grained model of capability antecedents and performance outcomes of servitization strategies. Our anchoring stance is the notion that the type of services offered by the servitizing unit – namely the implications for value creation and appropriation of BAS and ADS - is a key defining feature of its servitization strategy. We find that the adoption of BAS and ADS is driven by different capabilities: manufacturing capabilities impact the offering of BAS only, while service capabilities impact the offering of both BAS and ADS. We also find that: (i) BAS do not impact financial performance, but work as a platform for the offering of ADS; (ii) there seem to be naturally occurring servitization trajectories, involving the gradual development of balanced levels of BAS and ADS and the development of adequate levels of manufacturing and service capabilities; (iii) such 33

trajectories may enact a servitization paradox (Gebauer et al., 2005), by which performance declines for low levels of servitization maturity (low levels of ADS), but increases with significant levels of offering of ADS. Thus, servitization strategies based on the offering of BAS in isolation (product-centric business model) do not provide performance advantages and may be less sustainable in the long-term. The offering of ADS to a significant level, however, seems to represent a substantially different business model, exhibiting high levels of servitization maturity and generating performance advantages (Baines et al., 2010). Our study provides new insights into the servitization paradox and servitization trajectories. It contributes to the research on servitization business models (Kindström, 2010), servitization strategies (Eggert et al., 2014) and servitization trajectories (Oliva and Kallenberg, 2003; Gebauer et al., 2005; Martinez et al., 2010; Kowalkowski et al., 2015). Specifically, we propose a novel servitization trajectory vis a vis prior literature, based on using BAS as platform for ADS, rather than providing BAS first to a high extent. We highlight the need to consider both market breadth and depth in the analysis of servitization trajectories. Past research in these fields has produced several conjectures that received only scarce and partial empirical validation. Ours is one of the first studies to show empirically an explicit link between different servitization strategies, capabilities and servitization maturity. Moreover, it does so based on a large sample of manufacturing units from different countries and industrial sectors. Our findings also have important managerial implications, namely, by offering guidance for the design and deployment of servitization strategies. Specifically, our findings suggest that manufacturers wishing to servitize should distinguish between BAS and ADS and deploy a balanced adoption of BAS and ADS, using BAS as a platform. This 34

in turn, should be accompanied by the building of appropriate capabilities. Starting a servitization process through the offering of BAS requires both manufacturing and service capabilities, while moving to higher levels of servitization through the offering of ADS requires the development of strong service capabilities. Our study has some limitations, which open opportunities for future research. For the purpose of our study, it was appropriate to classify capabilities into the two categories of manufacturing and service. It would be important for future research to examine the capabilities at a more detailed level, distinguishing between different types of manufacturing and service capabilities (Ulaga and Reinartz, 2011), as well as encompassing external, network-based capabilities (Eloranta and Turunen, 2015). Moreover, although our model followed the literature indicating that service offerings were predicted by capabilities, some level of endogeneity (simultaneity) might also explain the association between these variables, i.e., capabilities explained service offerings, which in turn explained the further development of capabilities. Although some of these effects would be still captured in our trajectories model, longitudinal research obtaining data at multiple periods might provide a finer-grained analysis of these relationships. Finally, the findings on servitization trajectories are based on the observation of units at different stages of servitization, employing cross-sectional data. Future research should employ longitudinal designs to examine how individual manufacturing units implement servitization along time, as well as the extent to which deservitization movements may occur and why. We controlled for several important contextual factors of servitization (country, industry and business unit size). It would be important to further understand how other 35

contextual factors affect the adoption of different servitization strategies and trajectories, as well as how they affect their impact on performance (Sousa and Voss, 2008; Voss et al., 2016). Relevant factors could include the competitive environment (Neely, 2008), institutional factors (Turunen and Finne, 2014), industry technology lifecycle (Cusumano et al., 2015), position in the value chain (Baines et al., 2005) and managerial decision-making. Future research should also look at the impact of the advent of the internet of things and smart, connected products (Porter and Heppelman, 2014) on servitization strategies and trajectories. These technologies may increase efficiency in the provision of BAS by embedding them in products (e.g., remote monitoring and maintenance via sensors) (Wünderlich et al., 2015). They may also allow for the collection of rich, context-specific product usage data, increasing the depth and pace of learning about customers that is associated with bundles of smart products and BAS. Moreover, they may allow for some types of ADS to be provided in a more automated and efficient way, through the products themselves (e.g., products may autonomously learn and adapt to user preferences) (Porter and Heppelman, 2014). As a consequence, in the future, BAS and ADS may become more closely interconnected than at present.

References Baines, T., Higson, M. and Kay, G. (2005), “Strategic positioning - an integrated decision process for manufacturers”, International Journal of Operations & Production Management, Vol. 25 No. 2, pp. 180-201.

36

Baines, T. and Lightfoot, H. (2014), “Servitization of the manufacturing firm: Exploring the operations practices and technologies that deliver advanced services”, International Journal of Operations & Production Management, Vol. 34 No. 1, pp. 2-35. Baines, T., Lightfoot, H., Benedettini, O., Whitney, D. and Kay, J. (2010), “The adoption of servitization strategies by UK-based manufacturers”, Journal of Engineering Manufacture, Vol. 224 No. 5, pp. 815-829. Baines, T., Lightfoot, H., Smart, P. and Fletcher, S. (2013), “Servitization of manufacture: Exploring the deployment and skills of people critical to the delivery of advanced services”, Journal of Manufacturing Technology Management, Vol. 24 No. 4, pp. 637-646. Bowen, D., Siehl, C. and Schneider, B. (1989), “A framework for analyzing customer service orientations in manufacturing”, Academy of Management Review, Vol. 14 No. 1, pp. 75-95. Brax, S. (2005), “A manufacturer becoming service provider - challenges and a paradox”, Managing Service Quality, Vol. 15 No. 2, pp. 142-155. Brax, S., Jonsson, K. (2009), “Developing integrated solution offerings for remote diagnostics – A comparative case study of two manufacturers”, International Journal of Operations & Production Management, Vol. 29 No. 5, pp. 539-560. Cohen, J. (1988), Statistical Power Analysis for the Behavioral Sciences, Lawrence Erlbaum, Hillsdale, NJ. Cusumano, M. (2008), “The changing software business: Moving from products to services”, Computer, Vol. 41 No. 1, pp. 20-27. Cusumano, M., Kahl, S., Suarez, F. (2015), “Services, industry evolution, and the competitive strategies of product firms”, Strategic Management Journal, Vol. 36 No. 4, pp. 559–575.

37

da Silveira, G. (2011), “Our own translation box: exploring proximity antecedents and performance implications of customer co-design in manufacturing”, International Journal of Production Research, Vol. 49 No. 13, pp. 3833-3854. Davies, A. (2004), “Moving base into high-value integrated solutions: A value stream approach”, Industrial and Corporate Change, Vol. 13 No. 5, pp. 727-756. Dul, J., Hak, T., Goertz, G. and Voss, C. (2010), “Necessary condition hypotheses in operations management”, International Journal of Operations & Production Management, Vol. 30 No. 11, pp. 1170-1190. Eggert, A., Hogreve, J., Ulaga, W. and Muenkhoff, E. (2014), “Revenue and profit implications of industrial service strategies”, Journal of Service Research, Vol. 17 No. 1, pp. 23-39. Eisenmann, T., Parker, G. and Van Alstyne, M. (2011), “Platform envelopment”, Strategic Management Journal, Vol. 32 No. 12, pp. 1270-1285. Eloranta, V. and Turunen, T. (2015), “Seeking competitive advantage with service infusion: A systematic literature review”, Journal of Service Management, Vol. 26 No. 3, pp. 394-425. Ericsson, A. and Erixon, G. (1999), Controlling Design Variants: Modular Product Platforms. SME, Dearborn, MI. Fang, E., Palmatier, R. and Steenkamp, J. (2008), “Effect of service transition strategies on firm value,” Journal of Marketing, Vol. 72 No. 5, pp. 1-14. Fichman, M. and Cummings, J. (2003), “Multiple imputation for missing data: Making the most of what you know”, Organizational Research Methods, Vol. 6 No. 3, pp. 282-308. Galbraith, J. (2002), “Organizing to deliver solutions”, Organizational Dynamics, Vol. 31 No. 2, pp. 194-207. Gebauer, H., Fleisch, E. and Friedli, T. (2005), “Overcoming the service paradox in manufacturing companies”, European Management Journal, Vol. 23 No. 1, pp. 14-26.

38

Gebauer, H., Ren, G., Valtakoski, A. and Reynoso, J. (2012), “Service-driven manufacturing: Provision, evolution and financial impact of services in industrial firms”, Journal of Service Management, Vol. 23 No. 1, pp. 120-136. Golini, R. and Kalchschmidt, M. (2015), “Designing an expert system to support competitiveness through global sourcing”, International Journal of Production Research, Vol. 53 No. 13, pp. 3836-3855. Gorodnichenko, Y., Svejnar, J. and Terrell, K. (2010), “Globalization and innovation in emerging markets”, American Economic Journal: Macroeconomics, Vol. 2 No. 2, pp. 194-226. Gronroos, C. (2011), “Value co-creation in service logic: A critical analysis”, Marketing Theory, Vol. 11 No. 3, pp. 279-301. Hair, J., Black, W., Babin, B. and Anderson, R. (2010), Multivariate Data Analysis, Pearson, Upper Saddle River, NJ. Hair, J., Hult, G., Ringle, C. and Sarstedt, M. (2014), A Primer on Partial Least Squares Structural Equation Modeling (PLS-SEM), Sage, Thousand Oaks, CA. Hair, J., Ringle, C. and Sarstedt, M. (2011), “PLS-SEM: Indeed a silver bullet”, Journal of Marketing Theory and Practice, Vol. 199 No. 2, pp. 139-152. Helfat, C., Winter, S. (2011), “Untangling dynamic and operational capabilities: strategy for the (n)ever-changing world”, Strategic Management Journal, Vol. 32 No. 11, pp. 1243–1250. Honaker, J. and King, G. (2010), “What to do about missing values in time-series cross-section data”, American Journal of Political Science, Vol. 54 No. 2, pp. 561-581. Hong, P., Yang, M. and Dobrzykowski, D. (2014), “Strategic customer service orientation, lean manufacturing practices and performance outcomes”, Journal of Service Management, Vol. 25 No. 5, pp. 699-723.

39

IBM Corp. (2013), IBM SPSS Statistics for Windows, Version 22.0, IBM Corp., Armonk, NY. IMF (2015), World Economic Outlook Database, April 2015 Edition, available at: http://www.imf.org/external/pubs/ft/weo/2015/01/weodata/index.aspx (accessed November 2, 2015). IMSS

(2015),

The

International

Manufacturing

Strategy

Survey,

available

at:

http://www.manufacturingstrategy.net (accessed August 5, 2015). Kastalli, I. and van Looy, B. (2013), “Servitization: Disentangling the impact of service business model innovation on manufacturing firm performance”, Journal of Operations Management, Vol. 31 No. 4, pp. 169-180. Khan, O., Martin, C. and Creazza, A. (2012), “Aligning product design with the supply chain: A case study”, Supply Chain Management, Vol. 17 No. 3, pp. 323-336 Kindström, D. (2010), “Towards a service-based business model - Key aspects for future competitive advantage”, European Management Journal, Vol. 28 No. 6, pp. 479-490. Kindström, D. and Kowalkowski, C. (2009), “Development of industrial service offerings: a process framework”, Journal of Service Management, Vol. 20 No. 2, pp. 156-172. Kohtamaki, M., Partanen, J., Parida, V. and Wincent, J. (2013), “Non-linear relationship between industrial service offering and sales growth: The moderating role of network capabilities”, Industrial Marketing Management, Vol. 42 No. 1, pp. 1374-1385. Kuula, M. and Kauppi, K. (2015), “Performance impact of servitization under uncertainty”, in Proceedings of the 22nd EurOMA Conference, Neuchatel, Switzerland, June 26-July 1. Lay, G., Copani, G., Jäger, A. and Biege, S. (2010), “The relevance of service in European manufacturing industries”, Journal of Service Management, Vol. 21 No. 5, pp. 715-726.

40

Leachman, R. and Carmo, T. (1992), “On capacity modeling for production planning with alternative machine types”, IIE Transactions, Vol. 24 No. 4, pp. 62-72. Lin, Y. and Uhler, A. (2002), “Shortening the design for assembly process time for torque converter development”, Assembly Automation, Vol. 22 No. 3, pp. 248-259. Little, R. (1988), “A test of missing completely at random for multivariate data with missing values”, Journal of the American Statistical Association, Vol. 83 No. 404, pp. 1198-1202. Longoni, A. and Cagliano, R. (2015), “Environmental and social sustainability priorities”, International Journal of Operations & Production Management, Vol. 35 No. 2, pp. 216-245. MacKenzie, S., Podsakoff, P. and Jarvis, C. (2005), “The problem of measurement model misspecification in behavioral and organizational research and some recommended solutions”, Journal of Applied Psychology, Vol. 90 No. 4, pp. 710-730. Martinez V, Bastl, M., Kingston. J. and Evans S. (2010), “Challenges in transforming manufacturing organisations into product-service providers”, Journal of Manufacturing Technology Management, Vol. 21 No. 4, pp. 449-469. Mathieu, V. (2001a), “Product services: from a service supporting the product to a service supporting the client”, Journal of Business and Industrial Marketing, Vol. 16 No. 1, pp. 39-61. Mathieu, V. (2001b), “Service strategies within the manufacturing sector: Benefits, costs and partnership”, International Journal of Service Industry Management, Vol. 12 No. 5, pp. 451– 475. Mihm, J., Loch, C., Wilkinson, D. and Huberman, B. (2010), “Hierarchical structure and search in complex organizations”, Management Science, Vol. 56 No. 5, pp. 831-848. Miller, D., Hope, Q., Eisenstat, R., Foote, N. and Galbraith, J. (2002), “The problem of solutions: balancing clients and capabilities”, Business Horizons, Vol. 45 No. 2, pp. 3-12.

41

Nambisan, S. (2001), “Why service businesses are not product businesses”, MIT Sloan Management Review, Vol. 42 No. 4, pp. 72-80. Neely, A. (2008), “Exploring the financial consequences of the servitization of manufacturing”, Operations Management Research, Vol. 1 No. 2, pp. 103-118. Neill, S., Singh, G. and Pathak, R. (2014), “Technology and marketing capabilities in a developing economic context: assessing the resource-based view within a boundary condition”, International Journal of Business and Economics, Vol. 13 No. 1, pp. 75-92. Neu, W. and Brown, S. (2005), “Forming successful business-to-business services in goodsdominant firms”, Journal of Service Research, Vol. 8 No. 1, pp. 3-17. Oliva, R. and Kallenberg, R. (2003), “Managing the transition from products to services”, International Journal of Service Industry Management, Vol. 14 No. 2, pp. 160-172. Payne, A., Storbacka, K. and Frow, P. (2008), “Managing the co-creation of value”, Academy of Marketing Science, Vol. 36 No. 1, pp. 83-96. Podsakoff, P., MacKenzie, S., Jeong-Yeon, L. and Podsakoff, N. (2003), “Common method biases in behavioral research: a critical review of the literature and recommended remedies”, Journal of Applied Psychology, Vol. 88 No. 5, pp. 879-903. Porter, M., Heppelmann, J. (2014), “How smart, connected products are transforming competition”, Harvard Business Review, Vol. 92 No. 11, pp. 64-88. R Core Team (2015), R: A Language and Environment for Statistical Computing, R Foundation for Statistical Computing, Vienna, Available at https://www.R-project.org/. Sampson, S. and Froehle, C. (2006), “Foundations and implications of a proposed unified services theory”, Production and Operations Management, Vol. 15 No. 2, pp. 329-343. Sanchez, G. (2013), PLS Path Modeling with R, Trowchez Editions, Berkeley, available at: http://www.gastonsanchez.com/PLS_Path_Modeling_with_R.pdf. 42

Sanchez, G., Trinchera, L. and Russolillo, G. (2015), plspm: Tools for Partial Least Squares Path Modeling

(PLS-PM),

R

package

version

0.4.7,

available

at

http://CRAN.R-

project.org/package=plspm. Schmenner, R. (2004), “Service businesses and productivity”, Decision Sciences, Vol. 35 No. 3, pp. 333-347. Smith, L., Maull, R. and Ng, I. (2014),"Servitization and operations management: A service dominant-logic approach", International Journal of Operations & Production Management, Vol. 34 No. 2, pp. 242-269. Sousa, R. and Voss, C. (2008), “Contingency research in operations management practices”, Journal of Operations Management, Vol. 26 No. 6, pp. 697-713. Suarez, F., Cusumano, M. and Kahl, S. (2013), “Services and the business models of product firms: an empirical analysis of the software industry,” Management Science, Vol. 59 No. 2, pp. 420435. Tian, Y., Jia, Y., Sun, H., Li, D., Yang, Y. and Malik, S. (2012), “The moderating effect of service capability on the relationship between service delivery and business performance of manufacturing companies”, African Journal of Business Management, Vol. 6 No. 6, pp. 21692180. Timm, N. (2002), Applied Multivariate Analysis, Springer, New York. Tukker, A. (2004), “Eight types of product service system: eight ways to sustainability? Experiences from SusProNet”, Business Strategy & The Environment, Vol. 14 No. 4, pp. 246260. Turunen, T. and Finne, M. (2014), “The organizational environment’s impact on the servitization of manufacturers”, European Management Journal, Vol. 32 No. 4, pp. 603-615.

43

Ulaga, W., Reinartz, W. (2011), “Hybrid offerings: How manufacturing firms combine goods and services successfully”, Journal of Marketing, 75 (6), 5-23. Vandermerwe, S. and Rada, J. (1988), “Servitization of business: Adding value by adding services”, European Management Journal, Vol. 6 No. 4, pp. 314-324. Venkatraman, N. and Ramanujam, V. (1986), “Measurement of business performance in strategy research: a comparison of approaches”, Academy of Management Review, Vol. 11 No. 4, pp. 801-814. Von Nordenflycht, A. (2010), “What is a professional service firm? Toward a theory and taxonomy of knowledge-intensive firms”, Academy of Management Review, Vol. 35 No. 1, pp. 155-174. Voss, C., Perks, H., Sousa, R., Wittel, L., Wünderlich, N. (2016), “Reflections on context in service research”, Journal of Service Management, Vol. 27 No. 1, pp. 30-36. Wetzels, M., Odekerken-Schroder, G., and van Oppen, C. (2009). “Using PLS path modeling for assessing hierarchical construct models: guidelines and empirical illustration”, MIS Quarterly, Vol. 33 No. 1, pp. 177-195. Winter, S. (2003), “Understanding dynamic capabilities”, Strategic Management Journal, Vol. 24 No. 10, pp. 991–995. Wünderlich, N., Heinonen, K., Ostrom, A., Patrício, L., Sousa, R., Voss, C., Lemmink, J. (2015), ““Futurizing”

smart

service:

Implications

for

service

researchers

and

managers”,

Journal of Services Marketing, Vol. 29 No. 6/7, pp. 442-447. Zhang, Y. and Zhang, L. (2014), “Organizing complex engineering operations throughout the lifecycle”, Journal of Service Management, Vol. 25 No. 5, pp. 580-602. Zimmerman, D. (2004), “A note on preliminary tests of equality of variances”, British Journal of Mathematical and Statistical Psychology, Vol. 57 No. 1, pp. 173-181.

44

Table 1. Comparison between Basic Services and Advanced Services. Dimension Business model

Basic Services Product-centered (product provision and condition maintenance; value cocreation focuses on assisting the customer in realizing value potential embedded in product, by ensuring basic product operation)

Advanced Services Service-centered (co-creation of value in the customer’s specific context, beyond potential embedded in product)

Sample Studies Eggert et al, 2014; Kindström and Kowalkowski, 2009; Neely, 2008; Tukker, 2004; Smith et al., 2014; Mathieu, 2001a

Predominant contractual relationship

Transactional (selling products)

Relational (selling solutions through long-term relationships)

Martinez et al., 2010; Oliva and Kallenberg, 2003

Value added to customer through services

Low Reduced influence on the customers’ processes of value creation

High Significant influence on the customers’ processes of value creation

Gebauer et al., 2005; Mathieu, 2001a; Smith et al, 2014

Extent to which provider takes over customer processes related to the product

Low

Medium-High

Baines et al., 2013; Gebauer et al., 2005

Nature of the service processes

Quasi-manufacturing, simple, low contact, standardized

Professional, complex, high contact, customized

Baines et al., 2013; Gebauer et al., 2005; Martinez et al., 2010; Oliva and Kallenberg, 2003); Smith et al., 2014; Tukker, 2004; Mathieu, 2001a

Degree of customer interaction, involvement and co-creation

Low

High High number of customer touchpoints; a broad range of personnel are exposed to the customer

Baines et al., 2013; Martinez et al., 2010; Roels et al., 2010; Mathieu, 2001a

Competitive positioning

Services are not differentiated; hence, there is no deliberate service strategy

Services are differentiated and part of a deliberate service strategy

Gebauer et al., 2005

45

Table 2. Descriptive statistics (N = 763). xmin

xmax

𝑥

s

SCP1

1

5

3.01

1.17

SCP2

1

5

3.02

1.25

SCP3

1

5

3.07

1.11

MCP1

1

5

3.39

1.16

MCP2

1

5

3.67

1.18

MCP3

1

5

3.72

1.03

BAS1

1

5

3.18

1.45

BAS2

1

5

2.84

1.49

BAS3

1

5

3.40

1.36

ADS1

1

5

1.87

1.17

ADS2

1

5

2.58

1.36

ADS3

1

5

2.99

1.36

ADS4

1

5

2.83

1.34

ADS5

1

5

2.71

1.28

SALES

1

5

2.57

1.28

ROS

1

5

2.99

1.02

ISIC25

0

1

.30

.46

ISIC26

0

1

.13

.34

ISIC27

0

1

.17

.38

ISIC28

0

1

.25

.43

ISIC29

0

1

.10

.30

SIZE

3

150,000

2,912.01

12,361.71

1,496

101,169

32,635.71

24,881.29

SERVCAP

MANCAP

BAS

ADS

GDPCAP

46

Table 3. Measurement model estimates. Weight

Loading

Std. Error1

SERVCAP SCP1

.407

.885

.010

SCP2

.382

.791

.019

SCP3

.381

.885

.011

MANCAP MCP1

.377

.686

.040

MCP2

.496

.846

.022

MCP3

.389

.826

.023

BAS BAS1

.403

.854

.012

BAS2

.457

.869

.010

BAS3

.356

.727

.024

ADS

1

ADS1

.227

.641

.025

ADS2

.287

.782

.016

ADS3

.277

.770

.017

ADS4

.302

.828

.013

ADS5

.237

.705

.023

Cronbach’s

Composite

alpha

Reliability

.731

.814

.891

2.194 .549

.623

.695

.832

1.873 .719

.671

.753

.859

2.015 .633

.560

.801

.863

2.799 .737

AVE

Eigenvalues 1st

2nd

Loading standard errors based on 10,000 bootstrap replications.

47

Table 4. Pearson correlations. 1

2

3

4

5

6

7

8

9

10

11

12

1. ISIC25 2. ISIC26

-.256

3. ISIC27

-.298

-.178

4. ISIC28

-.380

-.227

-.264

5. ISIC29

-.216

-.129

-.150

-.192

6. LN(SIZE)

-.128

.056

-.033

-.025

.174

.156

-.215

-.062

.127

-.112

-.015

8. SERVCAP

-.173

.086

.075

.021

.040

.066

-.304

(.855)

9. MANCAP

-.201

.018

.054

-.004

.159

.131

-.118

.204

(.789)

10. BAS

-.224

.009

.031

.156

.035

.075

-.149

.433

.338

(.819)

11. ADS

-.245

.100

.095

.063

.015

.114

-.193

.526

.288

.674

(.748)

12. SALES

-.140

-.068

.039

-.008

.184

.582

.159

.112

.117

.085

.162

13. ROS

-.039

-.032

.047

.024

-.024

.075

-.024

.122

.011

.052

.164

7. LN(GDPCAP)

1AVE

.131

square roots (multi-item LVs) in main diagonal. N = 763. Absolute correlations ≥ .071 are significant (p < .05).

48

Table 5. Structural model estimates. Path

Estimate1

Std. Error2

Effect Size3

Hypothesis (direction)

MANCAP --> BAS

.247

.034

.076*

H1a (+)

MANCAP --> ADS

.036

.028

.003

H1b (n.s.)

SERVCAP --> BAS

.364

.033

.159**

H2a (+)

SERVCAP --> ADS

.269

.031

.116*

H2b (+)

BAS --> ADS

.531

.027

.436***

H3 (+)

BAS --> SALES

-.048

.040

.002

H4a (n.s.)

BAS --> ROS

-.115

.051

.007

H4b (-)

ADS --> SALES

.156

.041

.021*

H5a (+)

ADS --> ROS

.232

.050

.029*

H5b (+)

1

2

Significant estimates (p < .05) are in bold. Based on 10,000 bootstrap replications.

3

Effect sizes (Cohen, 1988): *** Large effect; ** medium effect; *small effect.

49

Table 6. ANOVA between clusters.

Basic services

Advanced services

Services investment

Services sales

Services priority

Low

High

Services

Services

Total

Welch

p-value

Mean

1.98

3.96

3.13

1769.69

< .001

S.D.

.71

.67

1.20

N

372

519

891

Mean

1.76

3.20

2.60

1034.92

< .001

S.D.

.60

.73

.98

N

372

519

891

Mean

4.35

7.69

6.31

25.82

< .001

S.D.

7.54

10.20

9.34

N

302

428

730

Mean

5.93

13.01

10.01

47.47

< .001

S.D.

15.01

14.32

15.03

N

356

483

839

Mean

3.40

3.98

3.74

78.97

< .001

S.D.

1.04

.77

.93

N

359

517

876

50

Manufacturing capability

H1a(+) H1b(n.s.)

H2b(+)

Sales

H4b(-)

H3(+)

H2a(+) Service capability

H4a(n.s.)

Basic services

Advanced services

H5a(+) H5b(+)

Profitability

Figure 1. Theoretical model.

Advanced Services (ADS) ADS=BAS

Servitized firm Naturally occurring servitization trajectory

Service capability

Basic Services (BAS) Manufacturing and Service capability

Figure 2. Capability development and servitization trajectories.

51

Figure 3. Scatterplot of basic and advanced services.

52